OPTIMIZING EXPLAINABLE RECOMMENDER SYSTEM USING KNOWLEDGE GRAPH AND LONG SHORT-TERM MEMORY (LSTM)-BASED DEEP LEARNING

Abstract

In today’s world, the prevalence of recommender systems (RS) has surged due to their ability to alleviate information overload by offering personalized content. Traditional RSs, categorized into content-based, collaborative filtering, and hybrid approaches, often lack transparency, leading to user distrust and dissatisfaction. To address this, we propose an LSTM-based deep learning model integrated with a knowledge graph for enhanced explainability and accuracy. Evaluated on MovieLens dataset, the model demonstrates superior performance across multiple metrics. Specifically, it achieves an accuracy of up to 92%, precision peaking at 68%, explainability scores ranging from 0.07 to 0.39, and diversity values between 1.07 and 1.28. Compared to state-of-the-art models like Wide & Deep, Matrix Factorization, and GRU4Rec, our approach consistently outperforms in predictive accuracy and interpretability. By leveraging temporal dynamics and structured knowledge, this framework addresses information sparsity challenges and provides meaningful insights into recommended items. These findings underscore the potential of deep learning techniques combined with knowledge graphs to advance explainable recommendation systems, enhancing both user satisfaction and system credibility.